The Press Release Was Right

On the eve of the Frankfurt Motor Show in September 2017, Volkswagen Group CEO Matthias Müller stood up and told the industry where electric cars were going. He called the plan Roadmap E. More than eighty new electrified models by 2025. More than twenty billion euros of direct investment through 2030. An electrified version of every one of the group’s roughly 300 models by 2030 at the latest. Volkswagen, restating a goal it had set the year before, intended to be the world’s number one in e-mobility by 2025.

Then he took his shot. The breakthrough, Müller said in the reported translation, “will not come from the self-styled pioneers. It will come from those who can get a new technology up and running on a truly relevant scale. We are not talking about a company with annual sales of 200,000 or 300,000 cars.” A month later, on a panel in Passau, he dropped the euphemism and named the numbers: Tesla sells about 80,000 cars against Volkswagen’s eleven million, destroys three-digit millions a quarter, and dismisses workers.

Everything in that presentation about the future was correct. That is the part worth sitting with.

It is tempting to file Volkswagen under the companies that laughed at the new thing. That is not what happened, and the truth is worse.

Volkswagen agreed with the forecast. It put a date on it, a number on it, and twenty billion euros behind it. It converted two German plants — Zwickau and Emden — into dedicated electric factories; Zwickau built its last combustion car in 2020. It built the platforms. It ran the ads. On the question of where the industry was heading, Volkswagen was not a skeptic. It was an evangelist with a budget.

On September 3, 2026, its supervisory board unanimously approved a plan calling for a further workforce adjustment of about 50,000 positions, beyond the roughly 50,000 already in existing programs. That is a target, not a completed act: implementation still has to be negotiated brand by brand, compulsory redundancies in Germany are excluded through the end of 2030, and the reductions run mostly through early retirement, attrition and voluntary severance. In June, Manager Magazin reported that Blume’s internal concept envisioned shedding up to 100,000 positions in total — roughly fifteen percent of a workforce of about 657,000.

The same decision acknowledged European overcapacity of more than 500,000 vehicles and stated that competitive follow-on production cannot currently be secured for the Emden, Zwickau, Hanover and Neckarsulm plants on a staggered basis from 2031 to 2034. Nothing has been closed; a production concept for Europe is due by the end of June 2027. But two of those four sites are Zwickau and Emden — the factories built for the future.

Volkswagen is not a corpse. It still leads Europe in electric vehicles and it still moves millions of cars. But it announced it would own the era, and instead it is cutting its model range by half and its product complexity by three quarters, conceding more than half a million units of European overcapacity, and negotiating which of its own factories survive the decade.

They were right about the future for nine straight years and lost it anyway.

An announcement is a decision you make once. A practice is a decision you make every morning. Volkswagen made the announcement and then spent the better part of a decade doing something other than the work — fighting over software, over plants, over cost structure, over who would eat the loss — while a competitor spent those same years shipping cars.

Look around your own industry right now and count the Roadmap Es. Nearly every company has an AI slide. Most have a committee, a pilot, a vendor, a policy, and a paragraph in the annual letter. Very few have anyone who uses the thing daily and can tell you what it is actually bad at. The slide is not the practice. The slide is the 2017 press release, and the 2017 press release was correct, and it did not save anybody.

Volkswagen’s failure was not a failure of foresight. It was a failure of doing. That is the more common failure, and it is the one that does not feel like failure while it is happening, because you have the memo and the memo is right.

Gal Gadot is starring in Doug Liman’s Bitcoin, alongside Casey Affleck, Pete Davidson and Isla Fisher. She has been blunt about it: “I am afraid of AI, but I am going to be afraid and look away? No. I am afraid. But I am playing with it and learning it and trying to educate myself.” And then the line that has been traveling: “The train has left the station. You’re going to work with it or be out of the game completely.”

Notice what she did not do. She did not announce a position. She did not publish a philosophy of AI, or declare herself a leader of the new thing, or make fun of anyone. She did three unglamorous things in order.

1. Learn

   She admits she is afraid of it and is educating herself anyway. Fear did not become a reason to stay ignorant. It became a reason to find out.

2. Do

   She made the film. Not a pilot, not a task force — a finished thing, shipped, with her name on it and her reputation exposed.

3. Understand

   Six months of contract to draw one line: sets and lighting, yes; her performance, no. That is not a slogan. Her lawyers were thorough enough that the actors’ guild called them for pointers.

Volkswagen had the announcement and skipped the doing. Gadot skipped the announcement and did the work. Nine years from now, one of those approaches will look obvious.

I would rather state the failure condition than hedge the claim.

Already true

The people getting the most out of these systems are not the ones with the best AI strategy. They are the ones who have shipped something with it and can tell you where it broke. That gap is visible now, in every company I walk into.

What has to happen

For the analogy to hold, the AI transition has to reward practice over posture the way the EV transition did. If it turns out that capital and distribution decide this one, the daily practitioner loses to the balance sheet and the memo was enough after all.

Where I am probably wrong

Volkswagen’s trouble is not purely a failure of doing. Chinese competition, tariffs, German labor costs and a co-determination structure that makes closing anything nearly impossible would have hurt a company that executed perfectly. And Gadot’s move is not available to most people — six months of lawyers is a luxury good. The lesson survives both caveats, but it is smaller than the headline makes it sound.

Do not underestimate this. Underestimating it does not look like laughing at it — that is the easy version, and almost nobody does that anymore. It looks like agreeing with it, loudly, in a deck, and never opening the tool. Learn it. Use it. Understand what you are actually protecting. Otherwise you get to be right about the future and demoralized inside of a decade, which is roughly where Volkswagen is sitting this week: a correct press release from 2017, and a plan to shed a hundred thousand jobs.

Sources

• Motoring Research — Roadmap E and the “self-styled pioneers” remarks, 11 September 2017. motoringresearch.com
• Green Car Congress — Roadmap E detail: model counts, investment, battery tender. greencarcongress.com
• The Drive — Müller’s Passau remarks on Tesla, October 2017. thedrive.com
• Volkswagen Group — Supervisory board approves Future Plan 2030, 3 September 2026. volkswagen-group.com
• electrive — Plants, overcapacity and the June 2027 deadline, 4 September 2026. electrive.com
• Reuters via The Star — Manager Magazin report on up to 100,000 positions, 26 June 2026. thestar.com.my
• Volkswagen Group — Four million all-electric vehicles delivered since 2013, 5 March 2026. volkswagen-group.com
• Best Selling Cars — BYD full-year 2025 sales and production data. best-selling-cars.com
• Variety — Gal Gadot on Bitcoin, the six-month contract, and the AI debate, September 2026. variety.com
• Consequence — “The train has left the station,” from the Happy Sad Confused interview, September 2026. consequence.net

Two things I left out because I could not confirm them. First, how the new 50,000 splits between voluntary departures and dismissals — Volkswagen has not published that, and about 20,000 of the earlier German tranche were signed voluntarily, so treat the breakdown as unavailable rather than as zero. Second, that AI was used only for sets and lighting on Bitcoin is Gadot’s characterization of her own production, not an independently verified fact. Neither gap changes the shape of the story, but both belong in it.